JadidJust-in AI Developments, Incidents & Deadlines
May 2026
SMTWTFS12345678910111213141516171819202122232425262728293031

AI news, Tuesday, May 5, 2026

7 stories

TLDR

  • OpenAI and Anthropic each announce competing PE-backed enterprise deployment JVs (May 4) — OpenAI's "The Deployment Company" ($10B, TPG/Brookfield/Bain) and Anthropic's JV ($1.5B, Blackstone/Goldman/H&F) both launch on the same day; the race to own the enterprise AI deployment layer is now a two-horse war.
  • Sierra AI raises $950M at $15.8B valuation (May 4) — $150M ARR in just 8 quarters; Fortune 50 customer base; enterprise agent demand shows no sign of slowing.
  • Meta acquires Assured Robot Intelligence (May 1) — Humanoid robotics team joins Meta Superintelligence Labs; physical AI is now a Big Tech priority alongside LLMsLarge language modelA neural network trained on huge amounts of text that generates text, answers questions and summarizes.Anthropic ↗.
  • GitHub Copilot shifts to usage-based billing June 1 (Apr 27) — Copilot code review will start consuming GitHub Actions minutes on private repos; 27 days to audit CI costs.
  • Mandiant: AI has industrialized cybercrime (M-Trends 2026, trending today) — TTE effectively negative; 28% of CVEsCommon Vulnerabilities and ExposuresA public catalog that gives each known security vulnerability a unique ID, a short description and at least one reference.NIST ↗ exploited within 24 hours of disclosure; median initial-access-to-handoff collapsed to 22 seconds.

Action Items

Things your engineering team should look at or do this week.

  • Audit GitHub Copilot code review costs before June 1 (27 days) — Starting June 1, every Copilot PR review on private repos consumes both AI Credits and GitHub Actions minutes under the new usage-based billing model. If you have automated Copilot code review triggers running in CI, open org Billing Settings → Actions now and use the preview bill GitHub is rolling out in early May to estimate projected overages. Public repos and self-hosted runners are unaffected. (GitHub Changelog | GitHub Blog)
  • Review enterprise AI vendor strategy given the new deployment JV landscape — Both OpenAI ("The Deployment Company") and Anthropic now have PE-backed "forward-deployed engineer" arms modeled on Palantir's approach. If your org is evaluating multi-year AI deployment contracts in the next 6 months, understand that these vehicles are designed for longer lock-in timelines and embedded teams — not just API keys. Read both announcements before entering any vendor negotiation. (TechCrunch)

APIs & Pricing

  • GitHub Copilot: usage-based (AI Credits) billing starts June 1 (announced Apr 27) — GitHub is replacing flat-rate plans with a credit model. Code review on private repos will consume both AI Credits and Actions minutes; public repos and self-hosted runners are unaffected. Opus 4.7 access is being restricted to Pro+ plans (Pro-tier plans lose it). GitHub is rolling out a preview bill tool in org billing settings in early May so you can project costs before the switch flips. (GitHub Changelog | GitHub Blog)

Security

  • Mandiant M-Trends 2026: AI industrializes cybercrime; time-to-exploit goes negative (report published Mar 24, resurging in coverage today) Trending — 28.3% of CVEs are now exploited within 24 hours of public disclosure — meaning active exploitation routinely precedes patches. Median time from initial access to credential handoff has collapsed from 8+ hours (2022) to 22 seconds (2025). Nation-states and financially motivated actors alike are using LLMs as force multipliers for hyper-personalized social engineering. New malware families PROMPTFLUX and PROMPTSTEAL actively query LLMs mid-execution to adapt and evade detection in real time. (Google Cloud Blog | The Hacker News | Help Net Security)

Industry & Business

  • OpenAI closes "The Deployment Company" — $10B PE-backed enterprise JV (May 4) Agents — Anchored by TPG, Brookfield, Advent, and Bain Capital ($4B in committed PE capital across 5 years), with OpenAI contributing up to $1.5B equity and retaining strategic control via super-voting shares. Unusually for a tech vehicle, it offers investors a guaranteed 17.5% annual return. The entity embeds OpenAI engineers directly inside PE portfolio companies in healthcare, logistics, manufacturing, and financial services — a Palantir-style "forward-deployed engineer" model at scale. (Bloomberg | The Next Web)

  • Anthropic launches $1.5B enterprise AI services JV with Blackstone, Goldman Sachs, Hellman & Friedman (May 4) — Announced on the same day as OpenAI's competing vehicle, Anthropic's JV also draws backing from General Atlantic, Apollo, GIC, and Sequoia. Initial targets are mid-market companies in healthcare, manufacturing, financial services, and real estate; Anthropic engineers embed directly inside client organizations rather than acting as traditional consultants. Fortune frames it as Anthropic taking direct aim at McKinsey, Accenture, and now OpenAI in the same motion. (Anthropic Blog | CNBC | Bloomberg)

  • Sierra AI raises $950M Series E at $15.8B post-money valuation (May 4) Agents — Led by Tiger Global and GV, with Benchmark, Sequoia, and Greenoaks. Sierra hit $150M ARR in 8 quarters — an unprecedented pace for enterprise SaaS — and counts Prudential, Cigna, Blue Cross Blue Shield, Rocket Mortgage, and one-in-three of the world's largest banks among its customers. CEO Bret Taylor (former Salesforce co-CEO) cited "intense demand" driven by the shift from chatbot pilots to production deployments at scale. (TechCrunch | CNBC)

  • Meta acquires Assured Robot Intelligence for humanoid AI push (May 1) Robotics/IoT — The San Diego-based startup, co-founded by UCSD professor Xiaolong Wang and Lerrel Pinto (whose previous robotics company was acquired by Amazon in March), joins Meta Superintelligence Labs. Meta is betting that the path to robust AGI runs through physical-world interaction, not language data alone; this is Meta's second robotics AI acquisition in two months. (TechCrunch | Bloomberg)

Policy & Safety

  • Colorado AI Act: June 30 compliance deadline in 56 days; rewrite still not enacted (Apr 2026) — Colorado's AI Act (SB 24-205) takes effect June 30, imposing anti-discrimination obligations on developers and deployers of "high-risk" AI in employment, housing, education, and financial services. A proposed rewrite from the March 2026 Work Group would delay enforcement to January 1, 2027 and streamline requirements — but it has not been enacted. The deadline is not a rumor; do not assume relief is coming. Teams deploying AI in Colorado-regulated contexts should continue preparing for June 30 as written. (Brownstein | Alston & Bird)

Type to search · ↑↓ to move · enter to open